Meshi D, Drew MR, Saxe M, Ansorge MS, David D, Santarelli L, Malapani C, Moore H, Hen R.
Hippocampal neurogenesis is not required for behavioral effects of environmental enrichment.
Nat Neurosci. 2006 Jun;9(6):729-31.
Please login to recommend the paper.
To make a comment you must login or register.
The interesting part of this story is that the authors can show that learning occurs following irradiation. This seems to indicate that learning can occur independently of hippocampal neurogenesis, but there are a few aspects of the study that make me favor a much more cautious interpretation of the results.
First, the behavioral data, as presented, raise some concerns. The latency-to-feed measure they use is not a standard measure of anxiety, and I would like to see this data backed up by either an elevated plus maze or open field experiments—these are far more common tests of anxiety in rodents. Second, the water maze data show that the enriched groups start off at noticeably lower levels than the standard groups, although the difference is not significant by their analysis. On day 5 the difference is barely significant, but, in fact, a regression through the learning curves for each group would not reveal a significant difference. All of the groups decrease their path length by about 700 cm in 5 days. It is not clear how the statistics for the last data points were conducted, either, because the degree of freedom suggest an n = 60 animals with only two groups (d.f. = 1, 60).
Their claim that neurogenesis might do different things in mice and rats is also very unusual (second-last paragraph). There is no evidence I know of to substantiate this speculation, but this statement may well need to be challenged by a more comprehensive study in the future.
Finally, our data (van Praag et al., 2002), suggest that new neurons take at least 5 weeks to become functional units in the CNS. This raises a curious point: Could new neurons actually be incorporated into and take part in a behavioral task that is only learned over 5 days? Probably not, though it might be claimed that an “existing pool” of new neurons might offer the means with which to learn more quickly. Obviously, this is not the case following the irradiation, which seems to have knocked out neurogenesis. In contrast, the original studies by Greenough and colleagues showed that environmental enrichment is also strongly associated with enhanced synaptogenesis. Our own studies (Eadie et al., 2005; Redila et al., 2006) show that both neurogenesis and synaptogenesis are increased by exercise, a major component of their enriched environment. Thus, it may be that this enhanced synaptogenesis is the main player in environmental enrichment’s benefits for learning and not neurogenesis. Unfortunately, the authors did not investigate this and leave us with a tantalizing conundrum that warrants a more involved investigation.
The new study by Meshi et al. convincingly demonstrates that hippocampal neurogenesis is not required for behavior improvements following environmental enrichment. The experimental design is exceptionally clean, and the work is beautifully done. In fact, the only significant hitch is that the results fly in the face of what we perhaps had expected these newborn neurons to do in the brain, and that this outcome is in direct opposition to the findings of a previously published study of enrichment-induced neurogenesis.
The previous study from Bruel-Jungerman et al., 2005 describes an essentially similar experiment to address the role of enrichment-associated neurogenesis in the hippocampus. Using a chemical antimitotic agent rather than X-irradiation to suppress hippocampal neurogenesis, Bruel-Jungerman et al. found that enrichment-associated improvements in long-term recognition behavior were eliminated in the treated mice. In contrast to Meshi et al., Bruel-Jungerman et al. concluded that enhanced neurogenesis resulting from enrichment is crucial for improved behavioral performance.
This set of papers is not the first instance of two seemingly similar experiments arriving at diametrically opposed results. Having been in the middle of one such conflict, I would suggest that rather than trying to identify who is right and who is wrong, we instead take the opportunity to recognize what the studies may reveal about the underlying biology. Given that both studies were carefully controlled and well designed, we might be able to refine our understanding of the significant variables and consequent outcomes through a careful comparison of the two methodologies.
There are several major differences between the experiments of Meshi and Bruel-Jungerman that may be significant in assessing these studies. The most obvious differences are in the species and gender of the animals under study: Meshi tested female mice, while Bruel-Jungerman used male rats. These differences may not be trivial: we currently understand very little about how species or gender interact with environment to affect outcome.
Second, Meshi used focal X-irradiation to ablate neural progenitors; Bruel-Jungerman used systemic methazoxy methanol acetate (MAM). Each method of suppressing cell division comes with its own benefits and limitations. X-irradiation allows nearly complete removal of the dividing progenitors and can be focally applied so that systemic effects are substantially reduced. However, irradiation is also associated with a prolonged inflammatory response in nearby tissue, and Meshi et al. found that the brain contained elevated CD68-positive microglia/macrophage staining for more than 1 month after irradiation treatment. Although the authors were careful not to begin their study until the overt signs of inflammation had resolved, the presence of prolonged microglial activation may change the local brain environment in ways we do not yet recognize.
In contrast, Bruel-Jungerman et al. used systemic MAM administration to knock down hippocampal progenitor proliferation during exposure to enrichment. The main advantages to MAM are that it doesn't require expensive equipment such as an x-ray machine, and that it is not known to induce an inflammatory response (although this has not been carefully examined). There are at least two main drawbacks to MAM. First, it is delivered systemically. While it does not grossly alter motor activity or general health, it likely has subacute effects on many systems that require active cell proliferation (gut, blood, etc.). Second, MAM interrupts proliferation by methylating DNA during cell division; one could imagine that it may also act to silence gene expression in postmitotic cells as chromatin opens for transcription. Gene transcription is required for the formation of new synapses, and this process may be interrupted along with neurogenesis in the enriched mice.
A third difference of note between the two studies is the measure each study used to assess what role the enrichment-associated neurogenesis had in learning and memory. Meshi et al. tested their mice in two behavioral tasks: the standard Morris water maze and a novelty-suppressed feeding test. Bruel-Jungerman et al. assessed animals using a novel-object recognition task. The three tasks vary greatly in nature, and likely tap different neural networks outside of the hippocampus. Comparing the results between studies, therefore, becomes almost impossible; instead, each must be assessed on its own terms. However, it should be noted that Bruel-Jungerman et al. observed the greatest differences between MAM-treated and saline-injected enriched animals at the longest retention intervals (48 hours). The Morris water maze has no such long-term memory requirement: mice are trained and tested over 9 consecutive days during which the task is rehearsed 3 times per day. It is possible that the addition of a final probe trial 24-48 hours after the last training session would reveal neurogenesis-dependent memory that was not needed to perform the task during the training period.
Enrichment is a variable protocol, and each study uses slightly different parameters to produce their enriched housing. In particular, Meshi et al. included exercise wheels, while Bruel-Jungerman et al. did not. Access to exercise may increase the extent of angiogenesis resulting from enrichment, and running alone was shown to be sufficient to reproduce the effects of full-scale enrichment on neuronal survival in the hippocampus (van Praag et al., 1999). The duration of enriched housing before behavioral testing also differed between the Meshi and Bruel-Jungerman studies. The longer enrichment period provided by Meshi et al. may have allowed for other functional and/or morphological changes associated with enrichment (synaptogenesis, increased LTP, greater dendritic branching, etc. (for review, see van Praag et al., 2000), to compensate for the loss of newborn neurons.
In summary, the studies by Meshi et al. and Bruel-Jungerman et al. show that even the most seemingly straightforward experiments are often more complicated than we realize. Biology is usually not simple, and sometimes it’s the messy conflicting data that will in time provide the most useful insight.
Activity level and respiration do co-vary when rodents play in an enriched environment. And Kheirandish et al. (2005) showed that hypoxia adversely affected working memory specially in male rats, and the dendritic branching and dopamine transport in the frontal cortex—not the hippocampus—of those male rats.
The implications of the above finding for Alzheimer's and depression can be extrapolated from this study. Thomas et al. (2006) found a decrease in serotonin transporter (a dopamine precursor) binding in the prefrontal cortex in Alzheimer disease subjects compared to both control and, ironically, depressed elderly, postmortem. They found no difference, however, in serotonin transporter binding between the depressed and the control subjects. That also held true when comparing Alzheimer disease subjects with and without depression. Serotonin transporter binding reduction does not increase in Alzheimer patients who also have major depression.